Skip to content Skip to navigation

Research & Ideas

Search this site

Steven H. Collins: New prosthetics should be better than the real thing

A mechanical engineer explains how more and better data is helping to create prosthetics unlike any before.

Person standing in a lab with a prosthetic leg on

Modern robotic technologies could help people with artificial limbs walk with less effort, more quickly and with better balance. | iStock/JohnnyGreig

For years prosthetic limbs were merely functional devices, but recent advances in robotics and neuroscience are transforming the very meaning of the word “prosthetic.”

Steve Collins is a mechanical engineer who is helping to lead that transformation to the benefit of people who’ve had an amputation, stroke or battlefield injury. The field has come a long way since the days of strap-on wooden legs.

Collins says that, rather than trying to merely mimic what the body does, he’s working on new ways of discovering prosthetic limb designs that outperform unimpaired bodies. His team uses advanced robotic systems that record and analyze the wearer’s response, continually tuning their mechanical assistance to optimize performance and make them better than ever before.

Join host Russ Altman and Steve Collins for a glimpse into the changing world of prosthetics on the latest episode The Future of Everything radio show. You can listen to The Future of Everything on Sirius XM Insight Channel 121iTunesGoogle PlaySoundCloudSpotifyStitcher or via Stanford Engineering Magazine.

Full Transcript

Russ Altman: Today on The Future of Everything, the future of prosthetics and exoskeletons. I have an uncle, Uncle Ron, who lost his leg when he was seven years old. It was in a driving accident, and in addition to wearing an artificial limb for his entire life, including to this day, as he’s around 80 years old, Uncle Ron devoted his life to the creation of artificial limbs to help others with the same challenges.

I remember him moving for a couple of years to the Fiji Islands where he trained folks in how to create, this was in the 60’s and 70’s, create artificial limbs, usually made out of wood. I remember his frequent visits to Vietnam, and other areas where war had left land mines, which continued to take their toll on civilians who accidentally set them off, and often lost legs.

I remember very primitive artificial limbs that he showed me in war in the 60’s and 70’s. They had cups for his thigh with padding, that were clearly there to try to make it a little bit more comfortable. I remember leather belts for attaching them as securely as possible to his remaining leg. Very crude bending at the knees. I can’t even remember If there was bending at the ankles, and there was always an issue of, and for a young boy this was what I remembered best, the matching shoes. Uncle Ron always had one shoe that he wore on his good leg, on his foot, and then he had the matching shoe that he would strap onto his artificial limb.

For a little kid, that was one of the most fascinating things in the whole kit and caboodle. And it was amazing to see my uncles facility with walking and activities of daily life. This really did go into the background, even though he must have had a discomfort, it was never apparent to me. And it was just my uncle. He did tell us about phantom limb pain, where he would have pain in parts of his leg that didn’t even exist anymore, and itching. And he said actually the itching was even worse than the pain in many ways. And this all points to the complex relationship between the brain, the limbs, and the nervous system.

Well, now we speed forward, and there’s been a development of new materials. We have robotics and software, we have methods for fabrication of prosthetics, and even exoskeletons. Which we’ll hear more about in a minute. And even these have advanced markedly. And there are expanded uses. Now we’re not just talking about the loss of a limb from trauma, but we now have patients with stroke and the inability to walk, and other neurological and muscular diseases.

Well Steve Collins, is a professor for Mechanical Engineering at Stanford. He has pioneered new methods for designing and deploying prosthetics and exoskeletons that simultaneously train the human in how to use these devices, as well as the device to kind of compliment the human’s capabilities. To harmonize and reach as good a functional capacity as possible. Steve, what do patients today experience when they need a prosthetic device, and how far have we come in terms of providing them useful devices that allow them to just function in their personal and professional lives without having this be the biggest deal?

Steve Collins: Sure, hearing about your uncle’s story is great, and I think a lot of things are, there’ve been improvements, but there are a lot of features that are really similar today. So people still have this padded formed, we call it a socket, that goes on the original limb.

Russ Altman: Socket, yes.

Steve Collins: Yep. And that provides some mechanical connection to the leg, and we now use carbon fiber mostly, and better polymers, but it’s still trying to do about the same thing. And we now use vacuum suspension to sort of suction cup action to keep if from sliding down.

Russ Altman: Yes, makes perfect sense.

Steve Collins: Instead of those leather straps which is nice.

Russ Altman: I do remember some of the woodwork on these legs was spectacularly beautiful, like when they made a fancy one, it was beautifully varnished and stained, and it really was like a little piece of art. Which may have fallen by the way side, but I do remember that they were spectacular.

Steve Collins: Oh yeah, that was a big component of early prosthetic limbs. One of the biggest suppliers was called Ohio Willow Wood. Advertising the kind of wood that they used in forming their limbs, we collaborated with them when I was a graduate student. But many people do, and I have a similar story from when I was a kid, I was much younger, but I met my great uncle, and I was a impetuous 8 year old or something, and my brothers and I asked him, what happened, how did you lose your leg? And he did this great thing, he said, “I’ll only answer one question. “I’ll tell you, but I’ll only answer one question.” We said, “Yeah, okay.” “It was bitten off.” And he left it right like that, which was fantastic that really stuck with me. Meeting him and learning about his experience.

Russ Altman: Was that true?

Steve Collins: No . I presume not, I don’t know. He only answered one question.

Russ Altman: He answered the one question.

Steve Collins: But actually no, I think he lost his limb to diabetes. And also, as it happened, I didn’t know this sister at the time, but he became really involved in research. And was a frequent participant in experiments at the University of Michigan, where I grew up. So anyhow, things have come a long ways in many dimensions. Especially with interfacing with the limb, but the basic function of the a prosthetic foot say, is still fairly similar to the device that you described. And many people are prescribed a solid ankle cushioned heel. It could be made of wood, or polymer. There’s some padding on the heel, and people who are more active will get something that’s shaped carbon fiber. So it’s more springy, but these are passive devices. So they don’t have the ability to do active work like our muscles can do, and they don’t have the ability to respond to our behavior and the environment, and change how they react to say, help you to power your way up stairs, or walk faster or slower, or recover from a little disturbance like a bump in the road. And those are the kinds of problems we’re trying to address.

Russ Altman: Yes, and it sounds like even not knowing very much about it, it seems like there are technologies that if appropriately used, could help in those areas. And I’m sure that’s what you’re focusing on.

Steve Collins: Yeah, absolutely. We think so, and so our group is one of, now maybe dozens of research groups across the world that are trying to incorporate these modern robotic technologies into prosthetic limbs that have more the function that we’re accustomed to from our biological limbs. So enable us to walk with less effort, and more quickly, and with better balance. Which are all big problems for people with amputation and stroke. In fact we think, that because of some of the advantages of robotic devices compared to the biological system, there are a lot of disadvantages, but there are some advantages. We think it’s possible to make robotic devices that enable people with disabilities to outperform people who haven’t had an injury like that.

Russ Altman: Right, so great, let’s do some definitional stuff, because I jumped right into it. But prosthetics, I think most people understand a prosthetic, I was describing the artificial legs, and there are arms. Tell me about exoskeletons, and then we’ll go to what the potential use cases are for these, and then we’ll talk about where we are in this process. So exoskeleton.

Steve Collins: Sure so, you can imagine something from a sci-fi movie, and that’s kinda the direction we’re headed. So the most popular vision of this would be the iron man suit.

Russ Altman: I was gonna ask about that, but I didn’t want to seem like a fool. So thank you for brazing it yourself.

Steve Collins: No, I think they capture the sort of, bold vision of exoskeletons that could, not only help recover some performance, but maybe extend your abilities beyond what comes naturally. And I guess this is on the continuum with the kinds of devices people are currently prescribed. Which we call orthosis. This is some molded plastic and metal that wraps around your existing limb, and tries to help say, provide more stiffness at the ankle joint to avoid say, dragging your toe if you have a drop foot or something like that. Or help keep the knee bending the way that we want it to bend during swing. The swing phase of walking, things like that. But there’re relatively basic technologies, and they can’t add power, they can’t respond to the environment. And so we’re trying to add some of these features that we think are really helpful in human performance.

Russ Altman: Got it. So I can think of the exoskeleton as kind of a descendant of the orthotic?

Steve Collins: Yes.

Russ Altman: Where we’re adding some energy and capabilities.

Steve Collins: A powered orthosis is what people called it originally.

Russ Altman: I’m sure the applications range from helping people who have had injury, but I’m sure it also could be in the kind of the Iron Man thing. Taking people who are kind of perfectly able to under normal observation, but giving them special industrial strength capabilities for industry or entertainment.

Steve Collins: Yeah that’s right.

Russ Altman: So are that full set of application kind of in range?

Steve Collins: Absolutely.

Russ Altman: In your work?

Steve Collins: Yes indeed. So the thing that gets me up in the morning is the idea that we can make devices to help with disabilities. That’s my biggest motivation.

Russ Altman: That’s the alpha.

Steve Collins: Absolutely, but it’s really hard, and so a lot of our early work has been with people who don’t have a disability. Able bodied folks. Mostly college age.

Russ Altman: Healthy volunteers.

Steve Collins: Healthy volunteers.

Russ Altman: They make the world go.

Steve Collins: Exactly, and this is where we’ve had our biggest breakthroughs and progress so far. And we do work with the military also with the Army, who are interested in preventing soldiers from getting injured when they carry lots of body armor out in the field.

Russ Altman: These packs can be 50 to 100 pounds is my understanding.

Steve Collins: Yeah, it’s ridiculous.

Russ Altman: And even though they’re young men, even though their young men, this can still take a toll.

Steve Collins: Oh yeah, absolutely. It’s a huge problem.

Russ Altman: This is The Future of Everything, I’m Russ Altman, I’m speaking with Professor Steve Collins about exoskeletons, orthotics and prosthetics. Okay so we have this full range. You’ve become well known of this idea of human in the loop, as a way to kind of implement a new orthotic and prosthetic technologies. Tell me what that’s all about, what’s the new idea there? It sounds like humans should have been in the loop all along, but maybe they weren’t.

Steve Collins: Yeah, well I guess they were, but we maybe weren’t really incorporating that into the design process. So this is a popular topic right now, and last year there were about 5,000 paper and patents in the area of powered prosthetic limbs and exoskeletons. And it’s been in the popular imagination for a long time. The media, that we’ve talked about, but to date there’s been only five or 10 devices that have demonstrated about that.

Russ Altman: I mean there was a guy who kicked a football at the world cup a couple years ago, and that got a lot of attention.

Steve Collins: It did absolutely, but...

Russ Altman: But almost proving the point that this was not really ready.

Steve Collins: Right, exactly, right yes. I think that’s a good illustration, right. So where we think the technology can be is much farther from where it is right now, and so we think that the reason that we’ve been making relatively slow progress, despite huge and vast but big groups of very smart people, is the design approach we’ve been using. Which has been to first use our intuition, or observations of some aspects of normal movement, or simple models to make a prediction as to what the device should do. And then we spend a few years making something that can embed that functionality, and then we finally get around to testing it on a person, and it doesn’t work. Almost all the time.

Russ Altman: And does not work. Can you unpack that a little bit for me?

Steve Collins: Sure.

Russ Altman: Is it, it doesn’t do as speced out?

Steve Collins: No.

Russ Altman: Or the patient hates it, or something in between?

Steve Collins: Yes, people often are very excited to try out the new technology, but after a while, maybe aren’t as thrilled about it. But no, it performed the mechanical, or robotic function that it’s designed to do.

Russ Altman: Right, ‘cause you guys are good, and you’re engineers you know how to build something that you’ve designed.

Steve Collins: Absolutely, we’re really good at that probably now. We’ve got fantastic technology, CAD and simulations, but what we’re really bad at, is predicting how the person is gonna respond. And so we think that is the crucial challenge I developing better assistive devices. Is we need to get a vaster assessment of how the person responds, and involve that response in the design process itself.

Russ Altman: Is the hypothesis that it will have to be at an individual level, that different people will have different needs, or do you at lest have some hope that if you involve them early, you’ll get a somewhat generic solution for multiple people?

Steve Collins: I think depending on the population, it could go either way, we just don’t know yet. So we’re trying to address both the problem of finding something that works for lots of people, and figuring out how to individualize something to customize it to an individual person’s needs.

Russ Altman: So can you tell me a story about kind of a story of a human in the loop kind of design effort that kind of worked or looks like it’s gonna work?

Steve Collins: Absolutely, so in our lab we developed some exoskeletons that help your ankle joint during walking, and for the first four or five years in the group we tried lots of different hand tuned controllers, and we saw modest benefits to the users improvements in the efficiency of walking that was our target, of about six percent.

Russ Altman: So the person has some sort of deficit in their ability of their ankle?

Steve Collins: Well these are again.

Russ Altman: These are normal.

Steve Collins: We’re starting with healthy controls.

Russ Altman: Okay, so you’re just trying to get them to walk more efficiently, so they could either do it longer or faster or whatever?

Steve Collins: Yes.

Russ Altman: Okay, so that’s the goal.

Steve Collins: Exactly, with the expectation it’s a good extent.

Russ Altman: And you were doing these tuning exercises which were not panning out so much?

Steve Collins: Not going great, no.

Russ Altman: Okay.

Steve Collins: We saw some small benefits, but much smaller than we thought possible.

Russ Altman: You said six percent.

Steve Collins: Six percent improvement energy.

Russ Altman: So just help me, would I recognize a six percent change in my efficiency of walking, or would that be in noise of like…

Steve Collins: Oh no you would. An example I like is the use of your arms in walking. So if you watch people walk, they have this characteristic arm swing right, that where your right arm moves with your left leg. And why do people do that? Well one reason seems to be to save energy. It actually saves you about eight percent of you metabolic energy.

Russ Altman: Now there’s a good number. So that helps me understand the six percent.

Steve Collins: Right, it sounds like a small number, but yet you walk people walk, they all swing their arms like this.

Russ Altman: And if you try to walk without doing that, which we’ve all done as a kid, it is very hard.

Steve Collins: Yeah, so if you put your hands in your pockets, you can do it, you look cooler, right. But you expend more energy. So you can watch people walking, and see how often aesthetics win out over economy. So anyways, these small differences do seem to affect peoples behavior five 10 percent.

Russ Altman: I’m sorry, so I derailed you.

Steve Collins: No.

Russ Altman: You were telling me the story, but so six percent is roughly the benefit you get from swinging your arms, little bit less?

Steve Collins: Right, compared to keeping them in your pockets.

Russ Altman: Gotcha, okay.

Steve Collins: But we thought we should really be able to do better than this. And that’s when we started developing new techniques for human in the loop optimization, where we would measure the persons energy use while they’re using the device, and then we’d feed that information back to this optimizers algorithm, that would then try different behaviors of the device. Different ways the device would move.

Russ Altman: Like tighten up a bolt, or give a little push at a certain point?

Steve Collins: Exactly. Like when does it push you, and how much does it push you. We talk about it in terms of these patterns of force, and torque at your joint. And so you walk with this optimizer, every couple minutes it would change the way the exoskeleton pushed on you. And after about an hour, we found that optimized assistance improved energy economy by 24 percent.

Russ Altman: So four times more.

Steve Collins: Four times bigger.

Russ Altman: And that makes perfect sense is that you’re now watching, you’re doing little mini experiments on a minute to minute basis. You’re getting feedback, and then you have the algorithms and obviously I’m sure you’re using AI, and the modern computation to say okay, we’re gonna tweak the variables. And that is a personalized kind of prescription, because if you put another person in that same, I’m guessing, tell me if I’m wrong, in that same device, they might benefit from the settings from the previous person, but you have a further ability to tweak. And is that what you’re seeing?

Steve Collins: That’s right. We’ve actually observed in an experiment, something that makes walking easier for me can make it harder for you. So it depends on the population and the kind of intervention, but you can use this process both to identify things that are generically good for lots of people, and ways to customize to an individual persons needs.

Russ Altman: And so that’s a great example. This is what you mean when you say human in the loop, and this sounds like it’s the future for, you design a generic solution, and then you tune it. Which raises a whole bunch of questions that we’ll get to in the next segment.

This is The Future of Everything, I’m Russ Altman, more with Dr. Steve Collins about prosthetics, orthotics, exoskeletons next on the SiriusXM Insight 121. Welcome back to The Future of Everything. I’m Russ Altman, I’m speaking with Steve Collins about exoskeletons and prosthetics.

So at the end of the previous segment, you made a really interesting comment that even for people who don’t have any disability, they have differences in what makes them more energetically efficient in walking. And that’s interesting cause I’ve heard also that somebody’s gate, the way that they walk, is almost like a fingerprint, that AI systems can use it to figure out if it’s me, or not me, because I swing my arms and my legs go out And I believe that, because I look at my kids, and some of my kids walk the way I do, and some of my other kids walk the way my wife does. So what is known about the individual dynamics and physics of walking, and how much variability there is, even in the normal population? And does at affect your work?

Steve Collins: Oh sure. Everyone’s different, everybody’s bodies different, and our nervous systems are different. So as you suggested, you can tell a person by their kinematics, by the way their body moves. And that affects how you need to assist them. And so we see that if we customize our devise to, even healthy controls, you get more of the benefit from the device. But the benefit’s even bigger in populations where there’s more heterogeneity. Like people who’ve had a stroke, it affects everyone really differently. Or an amputation, the places you might have pain, how much residual muscle you have, it’s different across everyone, and so we expect even bigger benefits if customization in these more diverse populations.

Russ Altman: And I don’t know if this question is too nerdy, but you talked about that 24 percent that you got to with the tuning. If I took that tune system and put it into somebody else, they probably wouldn’t get the full 24 percent. How much of it would they still benefit from the tuning that you did on person A for person B? Forgive the nerdy-ness.

Steve Collins: Oh not at all, no. That’s a great question. So that study I described to you was sort of this break out experiment for us, and very exciting for us. And we’ve since been following up with lot’s of additional experiments to probe these kinds of questions. And it looks like if you take the average of the best assistance profiles for 10 people, and you give it to a new person, they get a lot of the benefit. Maybe, on healthy controls, maybe 80 percent. Something like that.

Russ Altman: It’s the 80/20 rule again.

Steve Collins: Yeah exactly, yeah., totally. And so if you really wanna maximize performance you need to customize.

Russ Altman: But that 20 percent is worth while. It’s worth the extra effort perhaps.

Steve Collins: In some cases. So I think what we’re seeing is that if you wanna make a commercial product that’s widely available to keep costs down, and to make it easier for a person to use, maybe you wanna go for that 80 percent.

Russ Altman: Gotcha.

Steve Collins: So you do a lot of upfront development, and figure out what’s the, which this is still a hard question, what should the generic device look like. But once you have that information, then you can create these customized products, and then mass produce them, and keep the costs down, and actually reach a big market.

Russ Altman: Cool and I wanna end with, when are we gonna have this I our stores. But before that, tell me, this has been great, and we’ve jumped into a lot of things, but I wanna step back, what are the application areas that you see? You said that health drives you in the morning.

Steve Collins: Yes.

Russ Altman: What are the big areas where you see this can affect thousands or millions of people, and improve their life in terms of the application areas for this technology?

Steve Collins: Sure, so first we have people who really need a device like this, but it’s a smaller population. People with disabilities like amputation or stroke. We also have people who really need technology like this because of their occupation, like soldiers, other people who have to, you know, rescue workers, firefighters, carry heavy equipment, go long distances, need to keep going, high injury risk. And then for people that might use them for recreation. Like say runners, we have a collaboration with Nike, and you can imagine different ways in which this kind of device could improve peoples lives. Through consumer products, and that could reach a huge audience.

Russ Altman: Right, so there’s the three big groups, and within the health, I guess because of the aging population, stroke becomes a more and more important application. In terms of numbers, what parts of a healthcare have the biggest need for this?

Steve Collins: Right, stroke is a huge population. Millions of Americans per year are affected, and amputation is still hundreds of thousands of people have an amputation.

Russ Altman: And then I’m sure there are these dysmyelinating and neuromuscular diseases, you know, Multiple Sclerosis, and Lou Gehrig’s Disease where these people could also benefit greatly from these.

Steve Collins: Yes, it’s tens of millions of Americans that could benefit from that.

Russ Altman: This is The Future of Everything, I’m Russ Altman, I’m speaking with Steve Collins. So great, I wanted to get into those application areas. I know that you’ve done some work where the human is not in the loop, just robotic walking. How does that interact with the human in the loop part of the lab, is it that by understanding movement better, it’ll inform your work, or is it a separate interest? Where does that all fit in?

Steve Collins: Yeah, so actually my walking robot work was very early in my career. It was as an under graduate student. So I was at Cornell University, and I had this fantastic teacher, in a Dynamic course, name was Andrew Arina. And he showed this video of this passive dynamic walking robot. Said this is a system with no motors, no computers, and walking is just a natural movement of the systems like a pendulum swing.

Russ Altman: So it must have been down hill, am I right?

Steve Collins: Down hill yes, exactly.

Russ Altman: Because other otherwise we’re talking about perpetual motion.

Steve Collins: Yes.

Russ Altman: Okay, so down hill, but then no motors.

Steve Collins: But no motors, right. And so I got involved in this passive dynamic walking research, and then my big project there was to add a little bit of power to allow it to walk on level ground. Very, very simple primitive stuff, but it resulted in a really energy efficient walking robot. Similar cost to go a certain distance per unit in mass, as people. And that was a big thing.

Russ Altman: So can people Google this, cause I’m sure everybody wants to see this. If I did Google it, does it look familiar as a walk? Like is the solution that the robot found, or that you found, or that your mentor found, does that look humanoid, or is it an entirely different way to go forward?

Steve Collins: No it looks surprisingly natural. Like a person walking. At the time, we argued that it looked more natural than some of the powered robots at the time. Boston Dynamics Atlas looks pretty great these days, but this was 15 years ago. The passive movement of the body really it’s surprising how similar it looks to how people move their body, and it suggested, and still suggests that using those passive dynamics is part of what makes a person’s gait efficient.

Russ Altman: And it’s not hard, as somebody who’s a geneticist, it’s not hard for me to imagine that after a few hundred thousand, millions of years of evolution, we would’ve had a solution for walking that is extremely energetically efficient, and it sounds like you’ve shown that pretty much, almost empirically. That it’s a very good solution, if not optimal.

Steve Collins: Yes, there’s a big area of research in the field, and it looks like people are doing pretty well.

Russ Altman: Okay, so I wanted to ask about, so you’ve gotten us all worked up in a lather about these technologies, where are we in terms of having them be available for all of those three compelling application areas? How is the industry looking, what are the big challenges? What needs to happen?

Steve Collins: That’s a great question. So I think that the way’s you could see this technology getting out in the world are with the design and prescription, and training of these kinds of devices. So just coming up with what kinds of characteristics the devices should have. That’s a research and development thing. Some of it happens in academia, maybe some will happen in industry, and at the stage where you actually receive one of these devices, you can imagine kind of the way that you get your glasses prescribed. There’s this big system, and you try different settings, and see what works best. And so we developed these systems we call emulators. That are kind of like this VR system for physical devices that gives you feel of interacting with an exoskeleton or prosthetic limb without having to build it. And then you can make a few changes in software, and it feels like you’re using a different device. So you can imagine going to a clinic and strapping this on your leg, and then the prosthetist is tuning it.

Russ Altman: Yeah the eye glass analogy is a really compelling analogy, because that is a huge machine that turns that into a very simple pare of glasses, and you’re having a similar model now.

Steve Collins: Exactly, and we think this could result in people getting devices that are much more effective for them.

Russ Altman: And actually, again, really technical question, not even technical but, are these gonna be huge? These exoskeletons that you see now are these huge things that I would need a garage to keep it in. I’m sure that people in your field are think about what is the minimal size that we need to generate the forces, generate the motors, and be kind of compatible with life as human.

Steve Collins: Absolutely.

Russ Altman: And how is that going?

Steve Collins: That’s a really important direction for the field, and looking out five years, that’s our main focus. Is taking, once we know what these devices need to do, distilling that into a cheap and very energy efficient device so that lots of people could get access to it.

Russ Altman: So it sound like, in addition to all of this, you guys are fans of material science and power. Batteries, stuff like that, because you’re gonna need these things to be as miniature as possible, and as strong as possible. You know, carbon fiber, whatever. So this, it seems, would be a critical component of this emerging industry.

Steve Collins: Absolutely, we follow those developments very closely. And also, new actuation technologies. New ways of converting that energy stored in the battery into mechanical power on your leg.

Russ Altman: Because you don’t wanna take a hit in terms of the energy conversion from electricity to mode of force.

Steve Collins: Yes, electric motors are great, we can do a lot better, they’re still pretty heavy, and bulky.

Russ Altman: Thank you for listening to The Future of Everything, I’m Russ Altman, if you missed any of this episode, listen any time on demand with the SiriusXM app.